1. Field of the Invention
The present invention relates to computer techniques, and more particularly to thermal throttling duty estimation methods and systems for a central processing unit (CPU).
2. Description of the Related Art
An overheated central processing unit (CPU) may damage or cause a computer system to crash. To prevent this potential risk, computer systems are provided with CPU temperature monitoring and regulation functions.
CPU temperature reflects, and typically, is strongly related to CPU performance, that is, CPU temperature is driven higher when CPU performance increases. Conventionally, a computer system sets up a thermal throttle duty (TTD) and asserts temperature adjustment function when the computer system is booted. The TTD may be configured through a basic input/output system (BIOS).
Next, the computer system waits for and determines a signal THERMAL_L from the CPU. The signals from the CPU may reflect the need for temperature adjustment which may be done by increasing or decreasing CPU performance. When the signal is TRUE, the CPU is overheated, a timer is then triggered, and temperature adjustment starts. The timer counts for temperature adjustment time. When the timer counts up to a predetermined value, the temperature adjustment terminates. Thus, regulation of CPU temperature is achieved by CPU performance adjustment restricted by the temperature adjustment time counted by the timer.
The current method, however, has disadvantages. For example, the thermal throttle duty (TTD) remains unchanged from the time the system boots and thus lacks flexibility to match CPU temperature variation. As is known, regulation of CPU temperature is achieved by CPU performance adjustment. The CPU performance may be over-downgraded affecting system performance due to an improper fixed thermal throttle duty (TTD).
Hence, here is an important issue about how to dynamically adjust CPU performance with respect to CPU temperature to achieve proper temperature adjustment.